- Email: [email protected]
Recovery of High-Affinity Phage from a Nitrostreptavidin Matrix in Phage-Display Technology
ANALYTICAL BIOCHEMISTRY ARTICLE NO.
243, 264–269 (1996)
0515
Recovery of High-Affinity Phage from a Nitrostreptavidin Matrix in Phage-Display Technology Moshe Balass,*,† Ely Morag,* Edward A. Bayer,* Sara Fuchs,† Meir Wilchek,*,1 and Ephraim Katchalski-Katzir* *Department of Membrane Research and Biophysics, and †Department of Immunology, The Weizmann Institute of Science, 76100 Rehovot, Israel
Received July 25, 1996
A novel approach for the selection of high-affinity phage from phage–peptide libraries is described. The methodology employs a chemically modified form of streptavidin, termed nitrostreptavidin, which exhibits a reversible attraction for biotin. The new approach emulates conventional procedures in that a biotinylated probe, in this case biotinylated a-bungarotoxin, is attached to an immobilized streptavidin matrix. The phage library is introduced, and interacting phage particles are released under conventional acidic conditions (pH 2.2). At this stage, the primary peptide sequences characterizing the released phage are found to be identical with those previously known to interact with the toxin. However, other phage particles, which presumably interact more strongly than those released by acid, remain attached to the immobilized toxin. These can be released by virtue of the reversible biotin-binding properties of nitrostreptavidin. For this purpose, alkaline solutions (pH 10) or free biotin can be used. Using this approach, phage particles that recognize a-bungarotoxin were isolated; their peptide sequences were found to be similar to, but clearly distinct from, those collected by conventional acid elution. The affinity of the isolated phage was dramatically higher than that of phage obtained by the conventional methodology. In contrast, their synthetically prepared 15-mer peptides actually exhibited a lower affinity for the toxin than that shown by peptides prepared on the basis of the sequence obtained from conventional acid-eluted phage. This apparent discrepancy can be explained by an altered conformational state of the peptides in solution, compared to the epitopes expressed in situ on the phage surface. q 1996 Academic Press, Inc.
In conventional phage-display technology (1–3), streptavidin is commonly adsorbed onto an immobilizing matrix (e.g., microtiter plates, petri dishes) (2, 4– 1
To whom correspondence should be addressed.
7). Subsequently, a biotinylated form of a desired ‘‘target’’ or ‘‘selector’’ molecule (e.g., a toxin, antibody, or antigen) is attached, essentially through an irreversible interaction, to the immobilized streptavidin. The immobilized selector is then subjected to interaction with a phage-display library and interacting phage particles are thus bound to the solid phase. The phage particles are then released, usually by acid elution, which disrupts the phage–selector interaction. We were intrigued by the possibility that, after acid elution, a residual population of phage remains bound to the immobilized selector. If this is true, the remaining phage proteins might exhibit epitopes different from those of the acid-eluted phage. We assumed that release of a biotinyl probe from a streptavidin matrix would be difficult to achieve, and the question that follows is whether an isolated phage–target complex (i.e., phage–toxin, phage–antibody) would still be capable of replication. In order to sever the interaction between streptavidin and biotin, we recently modified the protein at the ortho position of the critical tyrosine residue in the biotin-binding site (8). One such streptavidin derivative, obtained using tetranitromethane as a modifying reagent, was a nitrotyrosine-containing streptavidin (i.e., nitrostreptavidin), which binds biotin efficiently under acidic or neutral conditions. However, the biotinbinding properties of nitrostreptavidin can be reversed either by high pH or by introduction of free biotin. Thus, biotinylated molecules bound to a nitrostreptavidin matrix can be released using either a buffer of pH 10 or a suitable biotin-containing buffer. This principle was illustrated in the present study by employing biotinyl a-bungarotoxin and nitrostreptavidin plates to screen a 15-mer phage–peptide library. Using this approach, we were able to demonstrate that high-affinity phage particles remain on the plates after conventional acid elution. The residual phage could be released using either alkaline buffers or
264
AID
0003-2697/96 $18.00 Copyright q 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.
AB 9840
/
6m23$$$141
11-11-96 18:02:14
abal
AP: Anal Bio
SELECTION OF HIGH-AFFINITY PHAGE
neutral buffers containing free biotin, and the resulting toxin-coated phage were capable of infecting Escherichia coli K91 cells. MATERIALS AND METHODS
Peptide-Display Library The 15-mer phage–peptide library employed in this study was provided by Chiron Corp. (Emeryville, CA) and was constructed by use of the phage M13 as described (3). This library consists of only about 2 1 107 different 15-residue peptide sequences, whereas theoretically, it can represent 3 1 1019 different 15-residue peptides. Each of the phage clones contains a 15-residue peptide fused to the N-terminus of the minor coat protein pIII, upon flanking with AG and PPPPPP at the N- and C-terminus, respectively. Preparation and Purification of Biotinylated a-Bungarotoxin (BTX)2 HPLC-purified BTX from the snake venom of Bungarus multicintus was purchased from Sigma Chemical Co. (St. Louis, MO; product No. T 3019). For biotinylation (7, 9), a solution of BTX (100 mg in 100 ml of 0.1 M NaHCO3 , pH 8.6) was incubated for 2 h at room temperature with 5 ml of a solution (1 mg/ml in dimethylformamide) of biotin amidocaproate N-hydroxysuccinimide ester (Sigma, B 2643). The reaction mixture was then dialyzed at 47C against phosphate-buffered saline (PBS; 0.14 M NaCl, 10 mM phosphate buffer, pH 7.4). Preparation of Nitrostreptavidin Nitrostreptavidin was prepared by chemical modification of the protein using tetranitromethane as recently described (8). Isolation of Peptide-Presenting Phage Selection and conventional acid elution of the attached phage particles was carried out as described previously (2, 7), using biotinylated BTX and streptavidin-coated petri dishes. In the new procedure, using nitrostreptavidin as a replacement for the native protein, phage were first eluted with acid, and the residually bound phage particles were subsequently released with biotin and/or at high pH. Experiments with nitrostreptavidin were performed as follows: A sample containing 1010 infectious phage particles was incubated overnight at 47C with biotinylated BTX [2.5 mg in 40 ml of 0.5% bovine serum albumin (BSA) in PBS]. The phage sample was diluted with 1 ml of 0.5% Tween 20 in PBS and allowed to react for
30 min at room temperature with gentle shaking in petri dishes (60 mm) coated with nitrostreptavidin (10 mg in 1 ml of 0.1 M NaHCO3 , pH 8.6). Unbound phage particles were removed by extensive washing (10 times, 10 min each) with 2 ml of PBS containing 0.5% Tween 20, and remaining phage were eluted with 0.8 ml of 0.1 M HCl titrated to pH 2.2 with glycine. Following acid elution, the dishes were washed again (four times), and residual, high-affinity phage were eluted using high concentrations of biotin (0.1 mg/ml, in either PBS, pH 7.4, or 0.1 M Na2CO3 , pH 10). The biotin-eluted phage particles were then amplified by infection of E. coli K91 cells and plating on LB plates containing 0.1 mg/ml ampicillin. The phage particles were purified and subjected to a second round of affinity purification with 1 mg of biotinylated BTX (in 40 ml of 0.5% BSA in PBS) and nitrostreptavidin-coated petri dishes. The acid-desorption and biotin-elution steps were repeated and the cycle (amplification, biotinylated BTX/nitrostreptavidin adsorption, acid desorption, biotin elution) was allowed to proceed for a third time, now using 0.1 mg of biotinylated BTX (in 40 ml of 0.5% BSA in PBS). After the third round, the biotin-eluted phage were screened for their specific binding to BTX using an enzyme-linked immunosorbent assay (ELISA), and the DNA of positive-reacting phage clones was isolated and sequenced (see below). ELISA of BTX Binding to Isolated Phage Wells of microtiter plates (Maxisorb, Nunc) were coated with 100 ml of a 1000-fold dilution (in 0.1 M NaHCO3 , pH 8.6) of rabbit anti-phage M13 serum by incubation overnight at 47C. Antibody-coated plates were washed three times with PBS containing 0.05% Tween 20, after which 100 ml of enriched phage clones (109 phage particles) were added to the wells and incubated for 1 h at 377C. Wells were then blocked for 1– 2 h at room temperature with a mixture of 1.5% BSA and 1.5% hemoglobin (final concentration of each in PBS), washed, and incubated overnight at 47C with biotinylated BTX (1 mg/ml, or as otherwise specified). For inhibition experiments, peptides were preincubated with biotinylated BTX for 30 min at 377C, prior to their addition to the phage-coated wells. After washing, bound biotinylated BTX was detected by incubation for 1 h at room temperature with a 2000-fold dilution of alkaline phosphatase-conjugated ExtrAvidin (BioMakor, Nes Ziona, Israel), using N-p-nitrophenyl phosphate (1 mg/ml in ethanolamine-buffered saline, pH 9; Sigma) as a substrate. The color (405 nm) that developed after about 1 h of incubation was determined using a microtiter plate reader. Inhibition of BTX Activity
2
Abbreviations used: BTX, a-bungarotoxin; PBS, phosphate-buffered saline; BSA, bovine serum albumin; ELISA, enzyme-linked immunosorbent assay; AChR, acetylcholine receptor.
AID
265
AB 9840
/
6m23$$$142
11-11-96 18:02:14
Inhibition by peptides of BTX binding to the acetylcholine receptor (AChR) was carried out in microtiter
abal
AP: Anal Bio
266
BALASS ET AL.
plates. Each well was coated overnight at 47C with 100 ml of purified Torpedo californica AChR (5 mg/ml in 0.1 M NaHCO3 , pH 8.6), and the procedure was continued as described above for the inhibition of biotinylated BTX binding to phage. Sequencing of Phage DNA Positive phage clones were propagated on E. coli K91 cells and grown on LB plates containing ampicillin (0.1 mg/ml). Phage DNA within the peptide-inserted region was sequenced using an automated 373 DNA sequencer and a Taq DyeDeoxy Terminator Cycle Sequencing Kit (Applied Biosystems, Perkin Elmer, Foster City, CA). Peptide Synthesis Peptide synthesis was carried out by the solid-phase method of Merrifield (10). RESULTS
Isolation of Peptide-Presenting Phage In a recent work (Balass et al., in preparation), we successfully employed biotinylated BTX for screening a 15-mer phage–peptide library. In that study, we were interested in identifying a peptide that mimics the BTX-binding site of receptor proteins, and hoped to produce a peptide that would block the biological activity of the toxin. In each biopanning cycle, an acidic solution (glycine–HCl, pH 2.2) was used to elute phage particles that were specifically immobilized via the target molecules (i.e., the biotinylated toxin) to streptavidin-coated petri dishes. In the present study, we used a modification of the conventional procedure in order to select for high-affinity phage. For this purpose, a chemically modified form of streptavidin (nitrostreptavidin) in which the binding-site tyrosine was converted to nitrotyrosine was used to produce a reversible biotin-binding affinity matrix (8). In this way, the nitrostreptavidin–biotin complex could be dissociated upon competition with free biotin, thereby eluting the phage and its associated biotinylated BTX. The eluted phage particles were amplified in E. coli cells and subjected to the next biopanning cycle. To select a high-affinity phage fraction, we first eluted the presumed lower-affinity phage particles with acid; during this step, biotinylated BTX remained on the plate and its linkage to the phage was disrupted. The remaining, noneluted, presumably higher-affinity phage particles were then dissociated from the nitrostreptavidin matrix using free biotin (see scheme in Fig. 1). The adsorbed phage were first washed four times with acid and then twice with biotin. The acid treatment from the first round of biopanning released most of the immobilized phage particles (Fig. 2). Subsequent
AID
AB 9840
/
6m23$$$142
11-11-96 18:02:14
FIG. 1. Schematic representation of biotin-induced and acid-induced dissociation sites between the phage and the nitrostreptavidin matrix. Acid dissociates the peptide–target interaction, whereas biotin displaces the biotinylated target molecule (together with the phage) from the nitrostreptavidin matrix.
washes with acid released negligible amounts of phage. Significantly, about 40% of the phage remained attached to the solid matrix, as indicated from subsequent biotin elution. Subsequent rounds served to amplify the number of selected phage. Following three successive rounds of biopanning, about 106 phage particles were obtained. ELISA analysis revealed that about 60% of the positive clones (n Å 90) from the third round of biopanning bound specifically to BTX. Comparison of BTX-Specific Sequences BTX-specific sequences obtained from biotin-eluted nitrostreptavidin matrices were compared with those obtained from acid-eluted phage. DNA from the eluted clones were sequenced, and the deduced 15-mer peptide sequences are shown in Table 1. The majority of clones, obtained by conventional acid elution from native streptavidin dishes (Table 1B), exhibited the sequence YMRYYESSLKYPDW. One of the 14 biotineluted phage (Table 1A) also displayed the same sequence; however, two new sequences were also obtained, both of which exhibited higher binding affinities than their acid-eluted counterparts. One of the latter sequences, EYMRYYESSLNPTRL, showed a stretch of nine residues that was identical with a portion of the major acid-eluted phage sequence, differing only in an additional acidic residue on its N-terminal side and the remaining five residues on the C-terminal portion. The biotin-eluted sequence (IWRYYEDSELMQPYR) obtained at the highest frequency (Ç70%) was also very similar to the other two sequences. Comparison of the high-affinity sequences suggested
abal
AP: Anal Bio
SELECTION OF HIGH-AFFINITY PHAGE
FIG. 2. Quantification of phage released by successive acid and biotin treatments during the first round of biopanning. Phage were bound to biotinylated BTX complexed to a nitrostreptavidin matrix. The initial acid elution (fraction 1) released about 10,000 low-affinity phage particles; much fewer were released following additional washes with acid (fractions 2–4). Incubation with biotin (fraction 5) then eluted another 4000 phage particles.
RYYESSL as a possible consensus motif. The N- and C-terminal portions of the peptides appear to be less restricted in their sequences, although several similar residues appear to characterize these regions. In this context, all of the sequences had at least one proline residue in the C-terminal region, ending with a hydrophobic (usually aromatic) residue. On the N-terminal portion, all had at least one aromatic and/or hy-
TABLE 1
Identification of Peptide Sequences Derived from BTX-Specific Phage Peptide sequencea
Frequencyb
A. Biotin elution IWRYYEDSELMQPYR EYMRYYESSLNPTRL YMRYYESSLKSYPDW
10 3 1
B. Acid elution YMRYYESSLKSYPDW FTYYQSSLEPLSPFY TMTFPENYYSERPYH PPPIFRYYEYWPTSY
16 3 2 1
Note. (A) Phage were eluted with free biotin (after acid elution) from nitrostreptavidin dishes, and (B) phage were desorbed from native streptavidin plates using acid solutions. a Amino acid sequences (deduced from DNA) of the 15-mer peptide inserted within the minor coat protein pIII. Peptide sequences are given in one-letter amino-acid codes. Residues included in a possible consensus sequence and a proline near the C-terminus are shown in bold. b Number of phage identified by random selection of BTX-positive phage.
AID
AB 9840
/
6m23$$$142
11-11-96 18:02:14
267
FIG. 3. Binding of BTX to biotin-eluted and acid-eluted phage. Curve A (l), biotin-eluted phage, bearing the sequence IWRYYEDSELMQPYR, selected by biopanning on nitrostreptavidin and elution by biotin (after extensive washing with acid). Curve B (s), acideluted phage (bearing the sequence YMRYYESSLKSYPDW) obtained by biopanning on native streptavidin and conventional elution with acid. Increasing concentrations of biotinylated BTX were added to the phage-coated wells in microtiter plates, and the extent of binding was determined using an avidin–enzyme conjugate as described in the text.
drophobic residue (usually a pair, e.g., IW, YM, or IF). The high-affinity motif, described in the present study, also bears a distinct resemblance to the sequence of the dodecapeptide [i.e., KHWVYYTCCPDT (11, 12)], derived from the putative cholinergic site of AChR from Torpedo (Balass, M., Katchalski-Katzir, E., and Fuchs, S., in preparation). Binding of BTX to both the acid- and the biotineluted phage was concentration-dependent (Fig. 3). The results clearly indicate that the affinity of the biotin-eluted phage was significantly higher than that of the acid-eluted phage. The binding assay was about 80-fold more sensitive for the biotin-eluted phage than for the acid-eluted phage. Thus, when displayed by the carrier protein or the intact phage in situ, the 15-mer sequences of the biotineluted phage indeed showed a higher affinity toward BTX than that exhibited by its acid-eluted counterpart. In contrast, when the 15-mer peptides themselves were prepared synthetically, their apparent in vitro affinity toward BTX was quite different (Fig. 4). The major acid-eluted peptide sequence was significantly more potent as a competitive inhibitor than either of the two high-affinity biotin-eluted phage. An IC50 value of less than 1006 M characterized the synthetic acid-eluted peptide, whereas values of about 5 1 1006 and ú1004 M were obtained for the two peptides derived from the biotin-eluted phage (i.e., EYMRYYESSLNPTRL and IWRYYEDSELMQPYR, respectively). This somewhat surprising result may well be related to the conformation of the peptide in its native state as part of a phage
abal
AP: Anal Bio
268
BALASS ET AL.
FIG. 4. Inhibition of BTX binding to Torpedo AChR by synthetic peptides. The two peptides, IWRYYEDSELMQPYR (l) and EYMRYYESSLNPTRL (j), were based on the selected sequences of the biotin-eluted phage (after extensive washing with acid) from the immobilized nitrostreptavidin. The peptide YMRYYESSLKSYPDW (m) was derived from acid-eluted phage. The peptide IWRKYEDSELMQPYR was used as a control (s), where K replaces Y at position 4. ELISA was performed as described under Materials and Methods; biotinylated BTX (50 ng/ml) was mixed with increasing concentrations of peptides, prior to their assay on AChR-adsorbed microtiter plates using an avidin–enzyme conjugate.
protein versus its less structured disposition in solution. DISCUSSION
In recent years, biopanning procedures used in phage-display technology have commonly featured biotinylated target molecules immobilized on streptavidin-coated matrices (microtiter plates, petri dishes, etc.). This procedure has provided the production of phage-display libraries with many of the advantages of avidin–biotin technology (13–15), including increased efficacy, versatility, and convenience of protocol. Before specifically adsorbed phage can be isolated, it is necessary to release them from the immobilized target molecule. This is usually accomplished by acid elution. A major flaw in this approach is the possibility that some of the phage, and perhaps the more interesting ones, remain attached to the immobilizing matrix because of their particularly high affinity for the target molecule. One recent attempt to overcome this problem and to isolate high-affinity phage has involved the in situ amplification of residual phage by introduction of bacteria to plates after the acid-elution step (16). A more appealing alternative would be to sever the bond between the target molecule and the matrix (see Fig. 1). However, immobilization of the target molecule by means of the near-irreversible streptavidin–biotin interaction would presumably preclude such an approach. The availability of a reversible form of avidin
AID
AB 9840
/
6m23$$$142
11-11-96 18:02:14
and streptavidin now allows us to consider this alternative. In this work, we applied a chemically modified form of streptavidin, nitrostreptavidin, in which the binding-site tyrosine was nitrated (8). Other reversible forms of streptavidin or avidin can also be used for this purpose, including recombinant mutant forms of either protein (17–19). This approach also demonstrates another important property of the phage-display system, namely, that phage replication is not incapacitated by the presence of the target molecule, either in solution or on the phage itself. This may seem surprising, since antibodies are known to inhibit the multiplication of haptenmodified phage—a principle that was widely employed during the early days of immunoassay development (20, 21). In the present system, however, there is only a limited number of peptide (three to five copies of protein pIII) per phage, and the phage–toxin complex is capable of replication under the conditions of the described protocol. Another interesting finding is that the in situ affinity of the peptide borne by the phage does not necessarily correlate with that in vitro (9, 22). When attached to the phage, the peptide comprises an exposed portion of a carrier protein (in this case, the minor coat protein pIII) and is ‘‘displayed’’ toward the medium in a conformationally relevant orientation. In contrast, synthetically prepared peptides in solution are less restricted in their conformation, and consequently the affinity toward their target molecule is impaired. Thus, in our system, the in vitro affinity of a lower-affinity (acideluted) peptide was superior to that of the high-affinity (biotin-eluted) phage. In this study, we clearly demonstrate that classical acid elution of phage is not complete and that residual high-affinity phage particles remain attached to the immobilizing matrix. Alternative measures for their release are therefore required, in which the target molecule is removed together with the phage in such a way that phage viability is not compromised. This principle was successfully demonstrated using a-bungarotoxin as a target for isolation of high-affinity peptides that mimic the acetylcholine receptor. Preliminary evidence in our laboratory (unpublished data) has also confirmed that other types of target molecules, notably monoclonal antibodies, can also be employed for this purpose. In all cases, the isolated phage particles were still capable of infection and proliferation. It is thus anticipated that the approach described here will prove to be generally applicable for the isolation of high-affinity peptides from phage-display libraries. ACKNOWLEDGMENTS The authors thank Chiron Corp. (Emeryville, CA) for the phagepeptide library used in this study. This work was supported by grants
abal
AP: Anal Bio
SELECTION OF HIGH-AFFINITY PHAGE from the Rashi Foundation and by a grant from the United States– Israel Binational Science Foundation (BSF), Jerusalem, Israel. Additional funding from the Baxter Healthcare Corporation (Chicago, IL) is gratefully acknowledged.
REFERENCES 1. Parmley, S. F., and Smith, G. P. (1988) Gene 73, 305–318. 2. Scott, J. K., and Smith, G. P. (1990) Science 249, 386–390. 3. Devlin, J. J., Panganiban, L. C., and Devlin, P. E. (1990) Science 249, 404–406. 4. Scott, J. K. (1992) Trends Biochem. Sci. 17, 241–245. 5. Clackson, T., and Wells, J. A. (1994) Trends Biotechnol. 12, 173– 184. 6. Cortese, R., Felici, F., Galfre, G., Luzzago, A., Monaci, P., and Nicosia, A. (1994) Trends Biotechnol. 12, 262–267. 7. Bayer, E. A., and Wilchek, M. (1990) Methods Enzymol. 184, 138–160. 8. Morag, E., Bayer, E. A., and Wilchek, M. (1996) Biochem. J. 316, 193–199. 9. Balass, M., Heldman, Y., Cabilly, S., Givol, D., KatchalskiKatzir, E., and Fuchs, S. (1993) Proc. Natl. Acad. Sci. USA 90, 10638–10642. 10. Merrifield, R. B. (1965) Science 150, 178–185.
AID
AB 9840
/
6m23$$$143
11-11-96 18:02:14
269
11. Neumann, D., Barchan, D., Safran, A., Gershoni, J. M., and Fuchs, S. (1986) Proc. Natl. Acad. Sci. USA 83, 3008–3011. 12. Conti-Tronconi, B. M., Tang, F., Diethelm, B. M., Spencer, S. R., Reinhardt-Maelicke, S., and Maelicke, A. (1990) Biochemistry 29, 6221–6230. 13. Bayer, E. A., and Wilchek, M. (1980) Methods Biochem. Anal. 26, 1–45. 14. Wilchek, M., and Bayer, E. A. (1988) Anal. Biochem. 171, 1–32. 15. Wilchek, M., and Bayer, E. A. (Eds.) (1990) in Methods in Enzymology, Vol. 184. Academic Press, San Diego. 16. Koivunen, E., Wang, B., and Ruoslahti, E. (1995) BioTechnology 13, 265–270. 17. Airenne, K. J., Sarkkinen, P., Punnonen, E.-L., and Kulomaa, M. S. (1994) Gene 144, 75–80. 18. Chilkoti, A., Tan, P. H., and Stayton, P. S. (1995) Proc. Natl. Acad. Sci. USA 92, 1754–1758. 19. Sano, T., and Cantor, C. R. (1995) Proc. Natl. Acad. Sci. USA 92, 3180–3184. 20. Becker, M. J., Conway-Jacobs, A., Wilchek, M., Haimovich, J., and Sela, M. (1970) Immunochemistry 7, 741–743. 21. Becker, J. M., and Wilchek, M. (1972) Biochim. Biophys. Acta 264, 165–170. 22. Yayon, A., Aviezer, D., Safran, M., Gross, J., Heldman, J., Cabilly, S., Givol, D., and Katchalski-Katzir, E. (1993) Proc. Natl. Acad. Sci. USA 90, 10643–10647.
abal
AP: Anal Bio
243, 264–269 (1996)
0515
Recovery of High-Affinity Phage from a Nitrostreptavidin Matrix in Phage-Display Technology Moshe Balass,*,† Ely Morag,* Edward A. Bayer,* Sara Fuchs,† Meir Wilchek,*,1 and Ephraim Katchalski-Katzir* *Department of Membrane Research and Biophysics, and †Department of Immunology, The Weizmann Institute of Science, 76100 Rehovot, Israel
Received July 25, 1996
A novel approach for the selection of high-affinity phage from phage–peptide libraries is described. The methodology employs a chemically modified form of streptavidin, termed nitrostreptavidin, which exhibits a reversible attraction for biotin. The new approach emulates conventional procedures in that a biotinylated probe, in this case biotinylated a-bungarotoxin, is attached to an immobilized streptavidin matrix. The phage library is introduced, and interacting phage particles are released under conventional acidic conditions (pH 2.2). At this stage, the primary peptide sequences characterizing the released phage are found to be identical with those previously known to interact with the toxin. However, other phage particles, which presumably interact more strongly than those released by acid, remain attached to the immobilized toxin. These can be released by virtue of the reversible biotin-binding properties of nitrostreptavidin. For this purpose, alkaline solutions (pH 10) or free biotin can be used. Using this approach, phage particles that recognize a-bungarotoxin were isolated; their peptide sequences were found to be similar to, but clearly distinct from, those collected by conventional acid elution. The affinity of the isolated phage was dramatically higher than that of phage obtained by the conventional methodology. In contrast, their synthetically prepared 15-mer peptides actually exhibited a lower affinity for the toxin than that shown by peptides prepared on the basis of the sequence obtained from conventional acid-eluted phage. This apparent discrepancy can be explained by an altered conformational state of the peptides in solution, compared to the epitopes expressed in situ on the phage surface. q 1996 Academic Press, Inc.
In conventional phage-display technology (1–3), streptavidin is commonly adsorbed onto an immobilizing matrix (e.g., microtiter plates, petri dishes) (2, 4– 1
To whom correspondence should be addressed.
7). Subsequently, a biotinylated form of a desired ‘‘target’’ or ‘‘selector’’ molecule (e.g., a toxin, antibody, or antigen) is attached, essentially through an irreversible interaction, to the immobilized streptavidin. The immobilized selector is then subjected to interaction with a phage-display library and interacting phage particles are thus bound to the solid phase. The phage particles are then released, usually by acid elution, which disrupts the phage–selector interaction. We were intrigued by the possibility that, after acid elution, a residual population of phage remains bound to the immobilized selector. If this is true, the remaining phage proteins might exhibit epitopes different from those of the acid-eluted phage. We assumed that release of a biotinyl probe from a streptavidin matrix would be difficult to achieve, and the question that follows is whether an isolated phage–target complex (i.e., phage–toxin, phage–antibody) would still be capable of replication. In order to sever the interaction between streptavidin and biotin, we recently modified the protein at the ortho position of the critical tyrosine residue in the biotin-binding site (8). One such streptavidin derivative, obtained using tetranitromethane as a modifying reagent, was a nitrotyrosine-containing streptavidin (i.e., nitrostreptavidin), which binds biotin efficiently under acidic or neutral conditions. However, the biotinbinding properties of nitrostreptavidin can be reversed either by high pH or by introduction of free biotin. Thus, biotinylated molecules bound to a nitrostreptavidin matrix can be released using either a buffer of pH 10 or a suitable biotin-containing buffer. This principle was illustrated in the present study by employing biotinyl a-bungarotoxin and nitrostreptavidin plates to screen a 15-mer phage–peptide library. Using this approach, we were able to demonstrate that high-affinity phage particles remain on the plates after conventional acid elution. The residual phage could be released using either alkaline buffers or
264
AID
0003-2697/96 $18.00 Copyright q 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.
AB 9840
/
6m23$$$141
11-11-96 18:02:14
abal
AP: Anal Bio
SELECTION OF HIGH-AFFINITY PHAGE
neutral buffers containing free biotin, and the resulting toxin-coated phage were capable of infecting Escherichia coli K91 cells. MATERIALS AND METHODS
Peptide-Display Library The 15-mer phage–peptide library employed in this study was provided by Chiron Corp. (Emeryville, CA) and was constructed by use of the phage M13 as described (3). This library consists of only about 2 1 107 different 15-residue peptide sequences, whereas theoretically, it can represent 3 1 1019 different 15-residue peptides. Each of the phage clones contains a 15-residue peptide fused to the N-terminus of the minor coat protein pIII, upon flanking with AG and PPPPPP at the N- and C-terminus, respectively. Preparation and Purification of Biotinylated a-Bungarotoxin (BTX)2 HPLC-purified BTX from the snake venom of Bungarus multicintus was purchased from Sigma Chemical Co. (St. Louis, MO; product No. T 3019). For biotinylation (7, 9), a solution of BTX (100 mg in 100 ml of 0.1 M NaHCO3 , pH 8.6) was incubated for 2 h at room temperature with 5 ml of a solution (1 mg/ml in dimethylformamide) of biotin amidocaproate N-hydroxysuccinimide ester (Sigma, B 2643). The reaction mixture was then dialyzed at 47C against phosphate-buffered saline (PBS; 0.14 M NaCl, 10 mM phosphate buffer, pH 7.4). Preparation of Nitrostreptavidin Nitrostreptavidin was prepared by chemical modification of the protein using tetranitromethane as recently described (8). Isolation of Peptide-Presenting Phage Selection and conventional acid elution of the attached phage particles was carried out as described previously (2, 7), using biotinylated BTX and streptavidin-coated petri dishes. In the new procedure, using nitrostreptavidin as a replacement for the native protein, phage were first eluted with acid, and the residually bound phage particles were subsequently released with biotin and/or at high pH. Experiments with nitrostreptavidin were performed as follows: A sample containing 1010 infectious phage particles was incubated overnight at 47C with biotinylated BTX [2.5 mg in 40 ml of 0.5% bovine serum albumin (BSA) in PBS]. The phage sample was diluted with 1 ml of 0.5% Tween 20 in PBS and allowed to react for
30 min at room temperature with gentle shaking in petri dishes (60 mm) coated with nitrostreptavidin (10 mg in 1 ml of 0.1 M NaHCO3 , pH 8.6). Unbound phage particles were removed by extensive washing (10 times, 10 min each) with 2 ml of PBS containing 0.5% Tween 20, and remaining phage were eluted with 0.8 ml of 0.1 M HCl titrated to pH 2.2 with glycine. Following acid elution, the dishes were washed again (four times), and residual, high-affinity phage were eluted using high concentrations of biotin (0.1 mg/ml, in either PBS, pH 7.4, or 0.1 M Na2CO3 , pH 10). The biotin-eluted phage particles were then amplified by infection of E. coli K91 cells and plating on LB plates containing 0.1 mg/ml ampicillin. The phage particles were purified and subjected to a second round of affinity purification with 1 mg of biotinylated BTX (in 40 ml of 0.5% BSA in PBS) and nitrostreptavidin-coated petri dishes. The acid-desorption and biotin-elution steps were repeated and the cycle (amplification, biotinylated BTX/nitrostreptavidin adsorption, acid desorption, biotin elution) was allowed to proceed for a third time, now using 0.1 mg of biotinylated BTX (in 40 ml of 0.5% BSA in PBS). After the third round, the biotin-eluted phage were screened for their specific binding to BTX using an enzyme-linked immunosorbent assay (ELISA), and the DNA of positive-reacting phage clones was isolated and sequenced (see below). ELISA of BTX Binding to Isolated Phage Wells of microtiter plates (Maxisorb, Nunc) were coated with 100 ml of a 1000-fold dilution (in 0.1 M NaHCO3 , pH 8.6) of rabbit anti-phage M13 serum by incubation overnight at 47C. Antibody-coated plates were washed three times with PBS containing 0.05% Tween 20, after which 100 ml of enriched phage clones (109 phage particles) were added to the wells and incubated for 1 h at 377C. Wells were then blocked for 1– 2 h at room temperature with a mixture of 1.5% BSA and 1.5% hemoglobin (final concentration of each in PBS), washed, and incubated overnight at 47C with biotinylated BTX (1 mg/ml, or as otherwise specified). For inhibition experiments, peptides were preincubated with biotinylated BTX for 30 min at 377C, prior to their addition to the phage-coated wells. After washing, bound biotinylated BTX was detected by incubation for 1 h at room temperature with a 2000-fold dilution of alkaline phosphatase-conjugated ExtrAvidin (BioMakor, Nes Ziona, Israel), using N-p-nitrophenyl phosphate (1 mg/ml in ethanolamine-buffered saline, pH 9; Sigma) as a substrate. The color (405 nm) that developed after about 1 h of incubation was determined using a microtiter plate reader. Inhibition of BTX Activity
2
Abbreviations used: BTX, a-bungarotoxin; PBS, phosphate-buffered saline; BSA, bovine serum albumin; ELISA, enzyme-linked immunosorbent assay; AChR, acetylcholine receptor.
AID
265
AB 9840
/
6m23$$$142
11-11-96 18:02:14
Inhibition by peptides of BTX binding to the acetylcholine receptor (AChR) was carried out in microtiter
abal
AP: Anal Bio
266
BALASS ET AL.
plates. Each well was coated overnight at 47C with 100 ml of purified Torpedo californica AChR (5 mg/ml in 0.1 M NaHCO3 , pH 8.6), and the procedure was continued as described above for the inhibition of biotinylated BTX binding to phage. Sequencing of Phage DNA Positive phage clones were propagated on E. coli K91 cells and grown on LB plates containing ampicillin (0.1 mg/ml). Phage DNA within the peptide-inserted region was sequenced using an automated 373 DNA sequencer and a Taq DyeDeoxy Terminator Cycle Sequencing Kit (Applied Biosystems, Perkin Elmer, Foster City, CA). Peptide Synthesis Peptide synthesis was carried out by the solid-phase method of Merrifield (10). RESULTS
Isolation of Peptide-Presenting Phage In a recent work (Balass et al., in preparation), we successfully employed biotinylated BTX for screening a 15-mer phage–peptide library. In that study, we were interested in identifying a peptide that mimics the BTX-binding site of receptor proteins, and hoped to produce a peptide that would block the biological activity of the toxin. In each biopanning cycle, an acidic solution (glycine–HCl, pH 2.2) was used to elute phage particles that were specifically immobilized via the target molecules (i.e., the biotinylated toxin) to streptavidin-coated petri dishes. In the present study, we used a modification of the conventional procedure in order to select for high-affinity phage. For this purpose, a chemically modified form of streptavidin (nitrostreptavidin) in which the binding-site tyrosine was converted to nitrotyrosine was used to produce a reversible biotin-binding affinity matrix (8). In this way, the nitrostreptavidin–biotin complex could be dissociated upon competition with free biotin, thereby eluting the phage and its associated biotinylated BTX. The eluted phage particles were amplified in E. coli cells and subjected to the next biopanning cycle. To select a high-affinity phage fraction, we first eluted the presumed lower-affinity phage particles with acid; during this step, biotinylated BTX remained on the plate and its linkage to the phage was disrupted. The remaining, noneluted, presumably higher-affinity phage particles were then dissociated from the nitrostreptavidin matrix using free biotin (see scheme in Fig. 1). The adsorbed phage were first washed four times with acid and then twice with biotin. The acid treatment from the first round of biopanning released most of the immobilized phage particles (Fig. 2). Subsequent
AID
AB 9840
/
6m23$$$142
11-11-96 18:02:14
FIG. 1. Schematic representation of biotin-induced and acid-induced dissociation sites between the phage and the nitrostreptavidin matrix. Acid dissociates the peptide–target interaction, whereas biotin displaces the biotinylated target molecule (together with the phage) from the nitrostreptavidin matrix.
washes with acid released negligible amounts of phage. Significantly, about 40% of the phage remained attached to the solid matrix, as indicated from subsequent biotin elution. Subsequent rounds served to amplify the number of selected phage. Following three successive rounds of biopanning, about 106 phage particles were obtained. ELISA analysis revealed that about 60% of the positive clones (n Å 90) from the third round of biopanning bound specifically to BTX. Comparison of BTX-Specific Sequences BTX-specific sequences obtained from biotin-eluted nitrostreptavidin matrices were compared with those obtained from acid-eluted phage. DNA from the eluted clones were sequenced, and the deduced 15-mer peptide sequences are shown in Table 1. The majority of clones, obtained by conventional acid elution from native streptavidin dishes (Table 1B), exhibited the sequence YMRYYESSLKYPDW. One of the 14 biotineluted phage (Table 1A) also displayed the same sequence; however, two new sequences were also obtained, both of which exhibited higher binding affinities than their acid-eluted counterparts. One of the latter sequences, EYMRYYESSLNPTRL, showed a stretch of nine residues that was identical with a portion of the major acid-eluted phage sequence, differing only in an additional acidic residue on its N-terminal side and the remaining five residues on the C-terminal portion. The biotin-eluted sequence (IWRYYEDSELMQPYR) obtained at the highest frequency (Ç70%) was also very similar to the other two sequences. Comparison of the high-affinity sequences suggested
abal
AP: Anal Bio
SELECTION OF HIGH-AFFINITY PHAGE
FIG. 2. Quantification of phage released by successive acid and biotin treatments during the first round of biopanning. Phage were bound to biotinylated BTX complexed to a nitrostreptavidin matrix. The initial acid elution (fraction 1) released about 10,000 low-affinity phage particles; much fewer were released following additional washes with acid (fractions 2–4). Incubation with biotin (fraction 5) then eluted another 4000 phage particles.
RYYESSL as a possible consensus motif. The N- and C-terminal portions of the peptides appear to be less restricted in their sequences, although several similar residues appear to characterize these regions. In this context, all of the sequences had at least one proline residue in the C-terminal region, ending with a hydrophobic (usually aromatic) residue. On the N-terminal portion, all had at least one aromatic and/or hy-
TABLE 1
Identification of Peptide Sequences Derived from BTX-Specific Phage Peptide sequencea
Frequencyb
A. Biotin elution IWRYYEDSELMQPYR EYMRYYESSLNPTRL YMRYYESSLKSYPDW
10 3 1
B. Acid elution YMRYYESSLKSYPDW FTYYQSSLEPLSPFY TMTFPENYYSERPYH PPPIFRYYEYWPTSY
16 3 2 1
Note. (A) Phage were eluted with free biotin (after acid elution) from nitrostreptavidin dishes, and (B) phage were desorbed from native streptavidin plates using acid solutions. a Amino acid sequences (deduced from DNA) of the 15-mer peptide inserted within the minor coat protein pIII. Peptide sequences are given in one-letter amino-acid codes. Residues included in a possible consensus sequence and a proline near the C-terminus are shown in bold. b Number of phage identified by random selection of BTX-positive phage.
AID
AB 9840
/
6m23$$$142
11-11-96 18:02:14
267
FIG. 3. Binding of BTX to biotin-eluted and acid-eluted phage. Curve A (l), biotin-eluted phage, bearing the sequence IWRYYEDSELMQPYR, selected by biopanning on nitrostreptavidin and elution by biotin (after extensive washing with acid). Curve B (s), acideluted phage (bearing the sequence YMRYYESSLKSYPDW) obtained by biopanning on native streptavidin and conventional elution with acid. Increasing concentrations of biotinylated BTX were added to the phage-coated wells in microtiter plates, and the extent of binding was determined using an avidin–enzyme conjugate as described in the text.
drophobic residue (usually a pair, e.g., IW, YM, or IF). The high-affinity motif, described in the present study, also bears a distinct resemblance to the sequence of the dodecapeptide [i.e., KHWVYYTCCPDT (11, 12)], derived from the putative cholinergic site of AChR from Torpedo (Balass, M., Katchalski-Katzir, E., and Fuchs, S., in preparation). Binding of BTX to both the acid- and the biotineluted phage was concentration-dependent (Fig. 3). The results clearly indicate that the affinity of the biotin-eluted phage was significantly higher than that of the acid-eluted phage. The binding assay was about 80-fold more sensitive for the biotin-eluted phage than for the acid-eluted phage. Thus, when displayed by the carrier protein or the intact phage in situ, the 15-mer sequences of the biotineluted phage indeed showed a higher affinity toward BTX than that exhibited by its acid-eluted counterpart. In contrast, when the 15-mer peptides themselves were prepared synthetically, their apparent in vitro affinity toward BTX was quite different (Fig. 4). The major acid-eluted peptide sequence was significantly more potent as a competitive inhibitor than either of the two high-affinity biotin-eluted phage. An IC50 value of less than 1006 M characterized the synthetic acid-eluted peptide, whereas values of about 5 1 1006 and ú1004 M were obtained for the two peptides derived from the biotin-eluted phage (i.e., EYMRYYESSLNPTRL and IWRYYEDSELMQPYR, respectively). This somewhat surprising result may well be related to the conformation of the peptide in its native state as part of a phage
abal
AP: Anal Bio
268
BALASS ET AL.
FIG. 4. Inhibition of BTX binding to Torpedo AChR by synthetic peptides. The two peptides, IWRYYEDSELMQPYR (l) and EYMRYYESSLNPTRL (j), were based on the selected sequences of the biotin-eluted phage (after extensive washing with acid) from the immobilized nitrostreptavidin. The peptide YMRYYESSLKSYPDW (m) was derived from acid-eluted phage. The peptide IWRKYEDSELMQPYR was used as a control (s), where K replaces Y at position 4. ELISA was performed as described under Materials and Methods; biotinylated BTX (50 ng/ml) was mixed with increasing concentrations of peptides, prior to their assay on AChR-adsorbed microtiter plates using an avidin–enzyme conjugate.
protein versus its less structured disposition in solution. DISCUSSION
In recent years, biopanning procedures used in phage-display technology have commonly featured biotinylated target molecules immobilized on streptavidin-coated matrices (microtiter plates, petri dishes, etc.). This procedure has provided the production of phage-display libraries with many of the advantages of avidin–biotin technology (13–15), including increased efficacy, versatility, and convenience of protocol. Before specifically adsorbed phage can be isolated, it is necessary to release them from the immobilized target molecule. This is usually accomplished by acid elution. A major flaw in this approach is the possibility that some of the phage, and perhaps the more interesting ones, remain attached to the immobilizing matrix because of their particularly high affinity for the target molecule. One recent attempt to overcome this problem and to isolate high-affinity phage has involved the in situ amplification of residual phage by introduction of bacteria to plates after the acid-elution step (16). A more appealing alternative would be to sever the bond between the target molecule and the matrix (see Fig. 1). However, immobilization of the target molecule by means of the near-irreversible streptavidin–biotin interaction would presumably preclude such an approach. The availability of a reversible form of avidin
AID
AB 9840
/
6m23$$$142
11-11-96 18:02:14
and streptavidin now allows us to consider this alternative. In this work, we applied a chemically modified form of streptavidin, nitrostreptavidin, in which the binding-site tyrosine was nitrated (8). Other reversible forms of streptavidin or avidin can also be used for this purpose, including recombinant mutant forms of either protein (17–19). This approach also demonstrates another important property of the phage-display system, namely, that phage replication is not incapacitated by the presence of the target molecule, either in solution or on the phage itself. This may seem surprising, since antibodies are known to inhibit the multiplication of haptenmodified phage—a principle that was widely employed during the early days of immunoassay development (20, 21). In the present system, however, there is only a limited number of peptide (three to five copies of protein pIII) per phage, and the phage–toxin complex is capable of replication under the conditions of the described protocol. Another interesting finding is that the in situ affinity of the peptide borne by the phage does not necessarily correlate with that in vitro (9, 22). When attached to the phage, the peptide comprises an exposed portion of a carrier protein (in this case, the minor coat protein pIII) and is ‘‘displayed’’ toward the medium in a conformationally relevant orientation. In contrast, synthetically prepared peptides in solution are less restricted in their conformation, and consequently the affinity toward their target molecule is impaired. Thus, in our system, the in vitro affinity of a lower-affinity (acideluted) peptide was superior to that of the high-affinity (biotin-eluted) phage. In this study, we clearly demonstrate that classical acid elution of phage is not complete and that residual high-affinity phage particles remain attached to the immobilizing matrix. Alternative measures for their release are therefore required, in which the target molecule is removed together with the phage in such a way that phage viability is not compromised. This principle was successfully demonstrated using a-bungarotoxin as a target for isolation of high-affinity peptides that mimic the acetylcholine receptor. Preliminary evidence in our laboratory (unpublished data) has also confirmed that other types of target molecules, notably monoclonal antibodies, can also be employed for this purpose. In all cases, the isolated phage particles were still capable of infection and proliferation. It is thus anticipated that the approach described here will prove to be generally applicable for the isolation of high-affinity peptides from phage-display libraries. ACKNOWLEDGMENTS The authors thank Chiron Corp. (Emeryville, CA) for the phagepeptide library used in this study. This work was supported by grants
abal
AP: Anal Bio
SELECTION OF HIGH-AFFINITY PHAGE from the Rashi Foundation and by a grant from the United States– Israel Binational Science Foundation (BSF), Jerusalem, Israel. Additional funding from the Baxter Healthcare Corporation (Chicago, IL) is gratefully acknowledged.
REFERENCES 1. Parmley, S. F., and Smith, G. P. (1988) Gene 73, 305–318. 2. Scott, J. K., and Smith, G. P. (1990) Science 249, 386–390. 3. Devlin, J. J., Panganiban, L. C., and Devlin, P. E. (1990) Science 249, 404–406. 4. Scott, J. K. (1992) Trends Biochem. Sci. 17, 241–245. 5. Clackson, T., and Wells, J. A. (1994) Trends Biotechnol. 12, 173– 184. 6. Cortese, R., Felici, F., Galfre, G., Luzzago, A., Monaci, P., and Nicosia, A. (1994) Trends Biotechnol. 12, 262–267. 7. Bayer, E. A., and Wilchek, M. (1990) Methods Enzymol. 184, 138–160. 8. Morag, E., Bayer, E. A., and Wilchek, M. (1996) Biochem. J. 316, 193–199. 9. Balass, M., Heldman, Y., Cabilly, S., Givol, D., KatchalskiKatzir, E., and Fuchs, S. (1993) Proc. Natl. Acad. Sci. USA 90, 10638–10642. 10. Merrifield, R. B. (1965) Science 150, 178–185.
AID
AB 9840
/
6m23$$$143
11-11-96 18:02:14
269
11. Neumann, D., Barchan, D., Safran, A., Gershoni, J. M., and Fuchs, S. (1986) Proc. Natl. Acad. Sci. USA 83, 3008–3011. 12. Conti-Tronconi, B. M., Tang, F., Diethelm, B. M., Spencer, S. R., Reinhardt-Maelicke, S., and Maelicke, A. (1990) Biochemistry 29, 6221–6230. 13. Bayer, E. A., and Wilchek, M. (1980) Methods Biochem. Anal. 26, 1–45. 14. Wilchek, M., and Bayer, E. A. (1988) Anal. Biochem. 171, 1–32. 15. Wilchek, M., and Bayer, E. A. (Eds.) (1990) in Methods in Enzymology, Vol. 184. Academic Press, San Diego. 16. Koivunen, E., Wang, B., and Ruoslahti, E. (1995) BioTechnology 13, 265–270. 17. Airenne, K. J., Sarkkinen, P., Punnonen, E.-L., and Kulomaa, M. S. (1994) Gene 144, 75–80. 18. Chilkoti, A., Tan, P. H., and Stayton, P. S. (1995) Proc. Natl. Acad. Sci. USA 92, 1754–1758. 19. Sano, T., and Cantor, C. R. (1995) Proc. Natl. Acad. Sci. USA 92, 3180–3184. 20. Becker, M. J., Conway-Jacobs, A., Wilchek, M., Haimovich, J., and Sela, M. (1970) Immunochemistry 7, 741–743. 21. Becker, J. M., and Wilchek, M. (1972) Biochim. Biophys. Acta 264, 165–170. 22. Yayon, A., Aviezer, D., Safran, M., Gross, J., Heldman, J., Cabilly, S., Givol, D., and Katchalski-Katzir, E. (1993) Proc. Natl. Acad. Sci. USA 90, 10643–10647.
abal
AP: Anal Bio